Enhanced Gradient Field Compensation in Multi-Channel Atomic Magnetometers with Adaptive Algorithms

Abstract

The utilization of spin-exchange relaxation-free (SERF) effects in atomic magnetometers is demonstrated significant potential for ultrasensitive biomagnetic field measurements. However, the presence of residual gradient fields could introduce additional spin-exchange relaxation, which will attenuate the response of the magnetometer. Consequently, it is crucial to compensate these residual gradients for maintaining high sensitivity and accuracy in SERF magnetometers. This study developed an optimized technique that leveraged both gradient and homogeneous coils to rapidly mitigate position-dependent gradients. By applying adaptive moment estimation (Adam) optimization, the quadratic loss function is efficiently solved and the optimal current is determined. This experiment demonstrated over 90% suppression of the residual fields and a significant reduction in the inter-channel deviations by an order of magnitude. In a simulated large gradient field environment, the sensitivities of multi-channel magnetometers improved from 100 to 15 $\mathrm{fT}/{\mathrm{Hz}}^{1/2}$ after compensation. The proposed method showcased substantial enhancements in residual homogeneity and sensor sensitivity, underscoring its efficiency. Compared with nested coil approaches, the coordinated use of gradient and homogeneous coils, along with Adam optimization, offered a robust, efficient, and rapid solution for gradient compensation. The proposed method effectively mitigated the adverse effects of residual fields and enhanced the performance of multi-channel SERF magnetometers. Importantly, this approach represents a more practical solution for gradient field compensation within a single cell, particularly for future commercial applications of array SERF magnetometers.